in all PI 3-kinase product binding PH domains (Y195 in .... (excluding the 5/ 6 loop) of ligand-free DAPP1-PH and DAPP1-PH with bound Ins(1,3,4,5)P4 is 0.8 AË .
Molecular Cell, Vol. 6, 373–384, August, 2000, Copyright 2000 by Cell Press
Structural Basis for Discrimination of 3-Phosphoinositides by Pleckstrin Homology Domains Kathryn M. Ferguson,* Jennifer M. Kavran,* Vijay G. Sankaran,* Emmanuel Fournier,† Steven J. Isakoff,† Edward Y. Skolnik,† and Mark A. Lemmon*‡ * Department of Biochemistry and Biophysics and The Johnson Foundation University of Pennsylvania School of Medicine Philadelphia, Pennsylvania 19104 †Department of Pharmacology New York University Medical Center Skirball Institute for Biomolecular Medicine New York, New York 10016
Summary Pleckstrin homology (PH) domains are protein modules of around 120 amino acids found in many proteins involved in cellular signaling. Certain PH domains drive signal-dependent membrane recruitment of their host proteins by binding strongly and specifically to lipid second messengers produced by agonist-stimulated phosphoinositide 3-kinases (PI 3-Ks). We describe X-ray crystal structures of two different PH domains bound to Ins(1,3,4,5)P4, the head group of the major PI 3-K product PtdIns(3,4,5)P3. One of these PH domains (from Grp1) is PtdIns(3,4,5)P3 specific, while the other (from DAPP1/PHISH) binds strongly to both PtdIns(3,4,5)P3 and its 5ⴕ-dephosphorylation product, PtdIns(3,4)P2. Comparison of the two structures provides an explanation for the distinct phosphoinositide specificities of the two PH domains and allows us to predict the 3-phosphoinositide selectivity of uncharacterized PH domains. Introduction Pleckstrin homology (PH) domains are protein modules of ⵑ120 amino acids found in many proteins involved in intracellular signaling, cytoskeletal organization, regulation of intracellular membrane transport, and modification of membrane phospholipids (Lemmon and Ferguson, 1998; Rebecchi and Scarlata, 1998). Most PH domains bind to phosphoinositides, and a few have been shown to drive membrane translocation of their host proteins through specific, high-affinity recognition of phosphoinositide head groups (Kavran et al., 1998). Experiments employing green fluorescent protein (GFP) fusions have demonstrated that certain isolated PH domains translocate to the plasma membrane of cells upon activation of phosphoinositide 3-kinases (PI 3-Ks). These include PH domains from Btk (Varnai et al., 1999), PKB (Watton and Downward, 1999; Gray et al., 1999), PLC-␥1 (Falasca et al., 1998), Grp1 (Venkateswarlu et al., 1998; Gray et al., 1999), Gab1 (Rodrigues et al., 2000), and EST684797/DAPP1/PHISH (Isakoff et al., 1998). Of these PH domains, some (e.g., from Btk and Grp1) ‡ To whom correspondence should be addressed (e-mail: mlemmon@
mail.med.upenn.edu).
recognize only phosphatidylinositol-(3,4,5)-trisphosphate (PtdIns(3,4,5)P3), while others (e.g., from PKB and DAPP1/ PHISH) bind similarly well to PtdIns(3,4,5)P3 and PtdIns (3,4)P2. Both lipid second messengers are almost undetectable in resting cells but accumulate following PI 3-kinase activation by almost all known cell surface receptors (Stephens et al., 1993; Vanhaesebroeck and Waterfield, 1999). PtdIns(3,4,5)P3 is the major direct product of agonist-dependent PI 3-kinases, and PtdIns(3,4)P2 is thought to arise largely through its 5-dephosphorylation by phosphatases such as SHIP, the SH2 domaincontaining inositol 5⬘-phosphatase (Stephens et al., 1993; Damen et al., 1996). Consistent with this, while PtdIns(3,4,5)P3 accumulation in stimulated platelets is immediate and transient, PtdIns(3,4)P2 accumulation is delayed and significantly more sustained (Franke et al., 1997). Therefore, a PH domain that recognizes only PtdIns(3,4,5)P3 will be recruited transiently to the plasma membrane, while one that binds both PtdIns(3,4,5)P3 and PtdIns(3,4)P2 will remain membrane associated for longer. Whether the PH domain of a protein recognizes PtdIns(3,4,5)P3 only, or both 3-phosphoinositides, will therefore have a significant impact on the time course of its response to PI 3-kinase activation. As described for extracellular signal-regulated kinase (ERK) pathways (Marshall, 1995), altering the duration of responses to extracellular stimuli can have profound effects on the nature of cellular consequences. There is also some evidence for separate control of cellular PtdIns(3,4)P2 and PtdIns(3,4,5)P3 levels. Activation of SHIPs alongside PI 3-kinase may modulate Tec kinase activation by altering the balance of PI 3-kinase product formation in favor of PtdIns(3,4)P2 (Bolland et al., 1998; Scharenberg et al., 1998). Furthermore, accumulation of PtdIns(3,4)P2 without preceding PtdIns(3,4,5)P3 production has been reported in platelets (Banfic et al., 1998) and oxidatively stressed fibroblasts (Van der Kaay et al., 1999). Whereas both the Grp1 and PKB PH domains translocate to the plasma membrane when fibroblasts are stimulated with PDGF, only the PKB PH domain translocates to the membrane when fibroblasts are oxidatively stressed, as PtdIns(3,4)P2 accumulates without PtdIns(3,4,5)P3 (Gray et al., 1999). With such growing evidence supporting distinct roles for PtdIns(3,4,5)P3 and PtdIns(3,4)P2 in cellular signaling, it is important to know which PH domains recognize which phosphoinositide(s). The PKB and Grp1 PH domains share a sequence motif that serves as a strong predictor of PI 3-kinase product binding (Isakoff et al., 1998), but it is not apparent from sequence analyses why Grp1-PH binds only PtdIns(3,4,5)P3 (Klarlund et al., 1997) whereas PKB-PH binds both PtdIns(3,4,5)P3 and PtdIns(3,4)P2. A recently cloned molecule named DAPP1 (Dowler et al., 1999) or PHISH (Rao et al., 1999) also contains a PH domain that binds both PtdIns(3,4)P2 and PtdIns(3,4,5)P3. The DAPP1/PHISH PH domain (encoded by EST684797) was first identified as a target of PI 3-kinase products in a yeast-based screen developed by Skolnik and colleagues (Isakoff et al., 1998). To understand the structural basis for their different phosphoinositide binding specificities, we have determined the X-ray crystal structures of the DAPP1/PHISH and Grp1 PH domains in complex with Ins(1,3,4,5)P4. Comparison
Molecular Cell 374
Table 1. Comparison of Inositol Phosphate-Binding Kd Values for Grp1 and DAPP1 PH Domains
Inositol trisphosphates Ins(1,3,4)P3 Ins(1,4,5)P3 Ins(1,5,6)P3 Ins(1,2,6)P3 Ins(3,5,6)P3 Inositol tetrakis-phosphates Ins(1,3,4,5)P4 Ins(1,3,4,6)P4 Ins(1,4,5,6)P4 Ins(3,4,5,6)P4 Ins(1,2,5,6)P4 Higher phosphates Ins(1,3,4,5,6)P5 InsP6
Grp1 PH (M)
DAPP1 PH (M)
⬎2.3 À4.5 0.13 À2.3 ⬇0.9
0.15 ⬇4.9 ⬎2 ⬎6.8 0.16
0.027 ⱖ1.4 ⱖ1.3 ⱖ0.9 0.18
0.043 0.64 ⬇2.2 0.87 0.29
0.33 ⬇5.8
0.19 ⬇2.4
Kd values for Ins(1,3,4,5)P4 binding by DAPP1-PH and Grp1-PH were determined by isothermal titration calorimetry (ITC), as described (Kavran et al., 1998). Competition by other inositol phosphates for binding of 3H-Ins(1,3,4,5)P4 to DAPP1-PH and Grp1-PH was then measured using a spin-column competition assay as described (Kavran et al., 1998). The Kd value reported in the table for each InsPn was estimated as (IC50(InsPn)/IC50(Ins(1,3,4,5)P4)) ⫻ Kd(Ins(1,3,4,5)P4). IC50’s were determined from mean competition curves generated from at least three repeats of each assay (see Kavran et al., 1998). Kd’s for Ins(1,3,4,5)P4 and for the inositol trisphosphates shown in Figure 1 are in italics.
of these structures provides insight into the determinants of binding specificity and has allowed us to predict correctly the selectivity of one as-yet-uncharacterized PH domain. Results and Discussion DAPP1-PH and Grp1-PH Recognize Different Features of PtdIns(3,4,5)P3 The PH domains from DAPP1 (DAPP1-PH) and Grp1 (Grp1-PH) both bind strongly to PtdIns(3,4,5)P3 and its soluble head group, D-myo-Ins(1,3,4,5)P4 (Table 1; all inositol phosphates are D-myo forms). In addition, DAPP1-PH can bind strongly to PtdIns(3,4)P2 and its head group, Ins(1,3,4)P3 (Kavran et al., 1998; Rao et al., 1999; Dowler et al., 1999), while Grp1-PH cannot (Klarlund et al., 1997; Kavran et al., 1998). This difference could result simply from DAPP1-PH being more promiscuous than Grp1-PH in inositol phosphate recognition or could reflect distinct specificities of the two PH domains. To determine which of these is true, we compared the specificities of DAPP1-PH and Grp1-PH using a 3 H-Ins(1,3,4,5)P4 competition assay (Kavran et al., 1998). Although the InsP4 binding properties of the two PH domains are very similar (Table 1), they show quite different InsP3 binding specificities (Figure 1). While DAPP1PH binds most strongly to Ins(1,3,4)P3, Grp1-PH instead prefers Ins(1,5,6)P3 (equivalent to Ins(3,4,5)P3 in phosphate group arrangement). These data argue that DAPP1PH focuses on the 1-, 3-, and 4-phosphates when binding to Ins(1,3,4,5)P4, while Grp1-PH focuses primarily on the 3-, 4-, and 5-phosphates. Of phosphoinsitides that occur physiologically, only PtdIns(3,4,5)P3 satisfies the phosphate group recognition requirements for Grp1-PH, while both PtdIns(3,4)P2 and PtdIns(3,4,5)P3 satisfy the
requirements for DAPP1-PH. In other words, the apparent promiscuity of DAPP1-PH in binding both PtdIns(3,4,5)P3 and PtdIns(3,4)P2 does not reflect a reduced degree of specificity but simply the fact that two physiological phosphoinositides happen to share the constellation of phosphate groups that it requires for high-affinity recognition. Determination of DAPP1-PH and Grp1-PH Crystal Structures To understand how DAPP1-PH and Grp1-PH recognize different patterns of head group phosphates, we used X-ray crystallography to determine the structures of the complexes that they form with Ins(1,3,4,5)P4. For DAPP1-PH, the structure of the unliganded PH domain was determined using multiple isomorphous replacement (MIR) and refined to 2.4 A˚ resolution. The resulting refined coordinates were used to solve the DAPP1-PH/ Ins(1,3,4,5)P4 structure to 1.8 A˚ resolution by molecular replacement (MR). For Grp1-PH, we were unable to crystallize the unliganded protein but obtained well-diffracting crystals of complexes with Ins(1,3,4,5)P4 and Ins(1,3,4,5,6)P5. We solved the structure of the Grp1PH/Ins(1,3,4,5,6)P5 complex to 1.9 A˚ resolution using MIR plus single-wavelength anomalous scattering (AS) from selenomethionine (SeMet)-containing crystals. An MR solution using the coordinates of the protein portion from this structure, together with AS phases from a SeMet-containing Grp1-PH/Ins(1,3,4,5)P4 crystal, was then used to solve the structure of Grp1-PH in complex with Ins(1,3,4,5)P4 to 2.5 A˚ resolution. Crystallographic details and results are summarized in Experimental Procedures and in Table 3. Structures of the PH Domains As expected, DAPP1-PH and Grp1-PH share the common PH domain fold, with a core 7-stranded -sandwich that is closed off at one splayed corner by the amphipathic C-terminal ␣ helix, and at the other splayed corner by the 1/2, 3/4, and 6/7 loops (Figure 2). DAPP1PH has no additional elements of secondary structure outside the basic PH domain core, resembling in this respect the N-terminal PH domain from pleckstrin (Yoon et al., 1994). Grp1-PH is structurally very similar in the core but contains an insertion (residues 334–352) in its 6/7 loop that forms two additional  strands (6⬘ and 6⬘⬘). Strands 6⬘ and 6⬘⬘ are colored red/brown in Figures 2C and 2D, and extend the antiparallel  sheet formed by strands 1 through 4. As described below, the extension of this  sheet appears to “deepen” the inositol phosphate binding site of Grp1-PH. DAPP1-PH and Grp1-PH also share the electrostatic sidedness that is characteristic of PH domains. Inositol Phosphate Binding to DAPP1-PH and Grp1-PH As shown in Figure 2, inositol phosphates bind to DAPP1-PH and Grp1-PH at the splayed corner of the PH domain that is formed by the 1/2, 3/4, and 6/7 loops. The ligand is bound at the center of the positively charged face of each PH domain and coincides with sites at which Ins(1,4,5)P3 binds to the PLC-␦1 PH domain (Ferguson et al., 1995) and Ins(1,3,4,5)P4 binds to the Btk PH domain (Baraldi et al., 1999). Figure 3 shows a close-up stereo view of Ins(1,3,4,5)P4 bound to DAPP1-PH and Grp1-PH, in which all direct hydrogen
PH Domain Recognition of 3-Phosphoinositides 375
Figure 1. DAPP1-PH and Grp1-PH Recognize Distinct Patterns of Phosphate Groups (A) Binding of Ins(1,3,4)P3 (magenta, squares), and Ins(1,5,6)P3 (which has the same phosphate arrangement as Ins(3,4,5)P3) (cyan, triangles) to DAPP1-PH (left) and Grp1-PH (right) is compared using a 3H-Ins(1,3,4,5)P4 competition assay (Kavran et al., 1998). DAPP1-PH binds 10-fold more strongly to Ins(1,3,4)P3 than to the Ins(3,4,5)P3 analog, while the converse is true for Grp1-PH. (B) Structures of the inositol trisphosphates. Ins(1,3,4)P3 (magenta) and Ins(3,4,5)P3 (cyan) are compared. Note that Ins(3,4,5)P3 is equivalent to Ins(1,5,6)P3 except in the positioning of the single axial hydroxyl group (2-OH). Ins(3,4,5)P3 is not commercially available, so Ins(1,5,6)P3 was used for the experiment shown in (A). Since Ins(3,5,6)P3 and Ins(1,3,4)P3 bind identically to DAPP1-PH (Table 1), we suggest that the axial 2-hydroxyl is not likely to be important in defining binding specificity.
bonds between the protein and ligand are marked. Despite the different specificities of these two PH domains, their Ins(1,3,4,5)P4 binding sites are very similar indeed. Approximately 75% of the side chain interactions with Ins(1,3,4,5)P4 in one complex have precise counterparts in the other. Superimposed on this framework of common interactions, the remaining 25% of the side chain hydrogen bonds are PH domain specific and must be responsible for the differences in ligand specificity. In the Grp1-PH/Ins(1,3,4,5)P4 complex, hydrogen bonding interactions are spread equally over the four phosphate groups of the ligand (Table 2). By contrast, in the complex formed between DAPP1-PH and Ins(1,3,4,5)P4 (which binds no more strongly than Ins(1,3,4)P3), there are fewer interactions with the 5-phosphate of Ins (1,3,4,5)P4 and more that involve the 4-phosphate. To illustrate the similarities and differences between the sets of interactions made in the two complexes, it is informative to consider each phosphate group in turn: 1-Phosphate The 1-phosphate of Ins(1,3,4,5)P4 forms hydrogen bonds with the side chains of T180 and (via water) K182 of DAPP1-PH. In the Grp1-PH complex, these interactions are reproduced with T280 and K282 (which interacts directly). 3-Phosphate The set of interactions made with the 3-phosphate of Ins(1,3,4,5)P4 is almost identical for DAPP1-PH and Grp1-PH. The 3-phosphate is “clamped” into position through the cooperation of a critical arginine side chain (R184 in DAPP1-PH; R284 in Grp1-PH) and the side chain amino group of a highly conserved lysine (K173 in DAPP1-PH; K273 in Grp1-PH). The third hydrogen bond donor for the 3-phosphate in each complex is a basic side chain that occupies the same spatial position, despite having quite different sequence locations in the two domains (K197 at the end of 3 in DAPP1-PH, and R305 at the beginning of 4 in Grp1-PH).
4-Phosphate DAPP1-PH forms more hydrogen bonds than Grp1-PH with the 4-phosphate group of Ins(1,3,4,5)P4 (Table 2). Common to both complexes are hydrogen bonds with the lysine in the center of Figure 3 that also contributes to “clamping” of the 3-phosphate (K173 in DAPP1-PH; K273 in Grp1-PH). Also common is a hydrogen bond to the side chain hydroxyl of a tyrosine that is conserved in all PI 3-kinase product binding PH domains (Y195 in DAPP1-PH; Y295 in Grp1-PH). Beyond these common interactions, the 4-phosphate of Ins(1,3,4,5)P4 forms three hydrogen bonds with two additional arginine side chains in DAPP1-PH that are listed in italics in Table 2 (R206 at the beginning of 4 and R235 in the 6/7 loop). Grp1-PH instead makes a single additional hydrogen bond with the side chain of H355 (from between 6⬘⬘ and 7). 5-Phosphate As might be expected from the specificities of the two PH domains, the 5-phosphate differs most in its hydrogen bonding when the DAPP1-PH and Grp1-PH complexes are compared. In both complexes there are four hydrogen bonds between the 5-phosphate of Ins(1,3,4,5)P4 and backbone amides at the beginning of the 1/2 loop (Table 2, bottom). However, while DAPP1-PH makes no side chain hydrogen bonds at all with the 5-phosphate (Figure 3A; Table 2, top), there are two side chain hydrogen bonds with the 5-phosphate group in the Grp1-PH complex (Figure 3B), which involve residues from the 6/7 insertion (K343 in 6⬘ and N354 in 6⬘⬘). Furthermore, the side chain of an arginine in the Grp1-PH 1/2 loop (R277) comes close (3.8 A˚) to hydrogen-bonding distance of the 5-phosphate. Structural Basis for the Ability to Distinguish between PtdIns(3,4,5)P3 and PtdIns(3,4)P2 This comparison of Ins(1,3,4,5)P4 binding by Grp1-PH and DAPP1-PH immediately suggests a structural basis
Molecular Cell 376
Figure 2. Comparison of DAPP1-PH and Grp1-PH Structures Ribbon representations of the DAPP1-PH and Grp1-PH structures. The C-terminal amphipathic ␣ helix, common to all PH domains, is colored blue.  strands that form the core  sandwich of the PH domains are colored green and are labeled 1–7. The phosphate positions in the bound inositol phosphates are numbered in red. N and C termini are labeled. (A) DAPP1-PH (aa 165–260) without bound Ins(1,3,4,5)P4. Two phosphate ions have been included in the model, which were placed in large peaks of electron density observed in the inositol phosphate binding site. The positions of these phosphates correspond to those of the 3and 4-phosphates of Ins(1,3,4,5)P4 in the structure of ligand-bound DAPP1-PH. (B) DAPP1-PH (aa 165–260) with bound Ins(1,3,4,5)P4. The right-hand view is from an aspect perpendicular to that on the left. Note the close correspondence of the backbone structures with and without (A) bound inositol phosphate. Very small changes are seen in the conformation of the 1/2 loop as result of Ins(1,3,4,5)P4 binding, as was also reported for Btk-PH (Baraldi et al., 1999). The position of the 5/6 loop is also altered, but as a result of differences in crystal packing contacts in the two structures. The rms deviation between 99 C␣ positions (excluding the 5/6 loop) of ligand-free DAPP1-PH and DAPP1-PH with bound Ins(1,3,4,5)P4 is 0.8 A˚. (C) Grp1-PH (aa 265–380) with bound Ins(1,3,4,5,6)P5. The two strands present in the insertion between 6 and 7 (labeled 6⬘ and 6⬘⬘—see text) are colored red/brown. The 6⬘ and 6⬘⬘ strands are close to the bound ligand and contribute side chain hydrogen bonds to the 4- and 5-phosphates. (D) Grp1-PH (aa 265–380) with bound Ins(1,3,4,5)P4. Two perpendicular aspects are shown. There is very good correspondence between the structures of Grp1-PH bound to Ins(1,3,4,5,6)P5 and Ins(1,3,4,5)P4. The only difference of note is that the 6-phosphate of Ins(1,3,4,5,6)P5 appears to “push out” the 1/2 loop slightly compared with its position in the Ins(1,3,4,5)P4 complex. The rms deviation between 122 C␣ positions of the two Grp1-PH complexes is 0.98 A˚, comparable to the difference between the two molecules in each asymmetric unit.
for the different specificities of the two PH domains. Above all, the fact that side chain hydrogen bonds to the 5-phosphate occur in the Grp1-PH/Ins(1,3,4,5)P4 complex, but not the DAPP1-PH/Ins(1,3,4,5)P4 complex, provides one explanation for the strong preference of Grp1-PH (but not DAPP1-PH) for PtdIns(3,4,5)P3 over PtdIns(3,4)P2. Further insight can be obtained by separating the inositol phosphate interactions into those that are common to the two PH domains and those that are PH domain specific. In the orientation used for the view of the binding site in Figures 4A and 4B, side chains shown on the right-hand side of the ligand are those that participate in interactions common to both PH domains, while those on the left are involved in domain-specific interactions. The side chain interactions common to both PH domains involve only the 1-, 3-, and 4-phosphates of Ins(1,3,4,5)P4, and their remarkable structural
similarity is illustrated by comparing the arrangement of T180, K182, R184, and K173 in DAPP1-PH (Figure 4A) with that of the equivalent side chains in Grp1-PH (T280, K282, R284, and K273 in Figure 4B). Beyond the Ins(1,3,4,5)P4 interactions that they share, both DAPP1-PH and Grp1-PH make an additional three unique side chain hydrogen bonds that cooperate with the common interactions to drive high-affinity Ins(1,3,4,5)P4 binding. These unique interactions must define the individual specificities of the PH domains and are listed in italic type in Table 2 (top). The labels for residues that participate in these unique interactions are boxed on the left-hand side of Figures 4A and 4B, and are colorcoded according to whether their side chains interact with the 4-phosphate (green) or 5-phosphate (blue) of Ins(1,3,4,5)P4. All interactions that are unique to the DAPP1PH/Ins(1,3,4,5)P4 complex are made with the 4-phosphate
PH Domain Recognition of 3-Phosphoinositides 377
Figure 3. Comparison of Ins(1,3,4,5)P4 Binding Sites in DAPP1-PH and Grp1-PH Stereo views of Ins(1,3,4,5)P4 in the binding site of (A) DAPP1-PH and (B) Grp1-PH. C␣ positions of the core  sandwich (plus C-terminal helix) of DAPP1-PH and Grp1-PH (molecule A) were overlaid using “O” and are viewed from identical orientations. All direct hydrogen bonds between the bound Ins(1,3,4,5)P4 and each PH domain are shown. The majority of water-mediated hydrogen bonds have been removed for clarity. A backbone C␣ trace is shown in light gray to highlight the remarkable similarity in structure of the two binding sites. The backbone is labeled according to the element of secondary structure to which it corresponds.
of Ins(1,3,4,5)P4 (through R206 and R235), while the interactions unique to the Grp1-PH/Ins(1,3,4,5)P4 complex involve both the 4-phosphate (H355) and 5-phosphate (K343 and N354) groups of the ligand. Surface representations (using the same orientation as in Figure 4) of the occupied binding sites on the two PH domains show that the 5-phosphate group is freely accessible to the solvent when Ins(1,3,4,5)P4 is bound to DAPP1-PH (Figure 5A) but is nestled within a clearly defined pocket or cavity on the PH domain surface when bound to Grp1-PH (Figure 5B). This pocket is formed by the juxtaposition of side chains from the 1/2 loop and the 6/7 insertion in Grp1-PH and has an area of positive electrostatic potential at its opening contributed by the K343 and R277 side chains (below and to the left of the 5-phosphate, respectively, in Figure 5B). There is no equivalent region of positive electrostatic potential close to the 5-phosphate in the DAPP1-PH/ Ins(1,3,4,5)P4 complex, since K343 has no counterpart in this PH domain, and R277 of Grp1-PH is replaced by a leucine (L177) in DAPP1-PH (Figure 6). Calculations
of solvent-accessible surface area show that the 3- and 4- phosphates of Ins(1,3,4,5)P4 are both 90%–100% buried in the complexes formed with DAPP1-PH or Grp1-PH. The 5-phosphate is also 90% buried in the Grp1-PH/Ins(1,3,4,5)P4 complex but is only 56% buried in the DAPP1-PH/Ins(1,3,4,5)P4 complex. The 1-phosphate group of Ins(1,3,4,5)P4 remains ⵑ60% accessible in both complexes, sufficiently accessible such that it will accommodate the diacylglycerol moiety of the inositol phospholipids to which the PH domains bind. From the comparison of ligand binding by these two PH domains, it can be argued that DAPP1-PH binds strongly to both PtdIns(3,4,5)P3 and PtdIns(3,4)P2 since it is able to make an equal number (14) of side chain hydrogen bonds with the 1-, 3-, and 4-phosphates of either ligand (there are none to the 5-phosphate). Grp1-PH can make only 12 of these hydrogen bonds with PtdIns(3,4)P2 (to which it binds poorly) but is able to make two additional unique hydrogen bonds to the 5-phosphate of PtdIns(3,4,5)P3, to which it binds strongly. This pair of additional hydrogen bonds (Table
Molecular Cell 378
Table 2. Direct Side Chain Hydrogen Bonds to Ins(1,3,4,5)P4 and Backbone Hydrogen Bonds to Ins(1,3,4,5)P4 DAPP1-PH
Grp1-PH
Btk-PH
IP4
Residue
Atom
Distance
Residue
Atom
Distance
Residue
Atom
Distance
O1 O11 O11 O12
Thr-180 Thr-180
O␥1 O␥1
2.9 3.0
Thr-280 Thr-280 Lys-282
O␥1 O␥1 N
2.8 3.1 3.4
Asn-24
O␦1
3.1
Lys-26
N␦
3.4
O2
Thr-180
O ␥1
3.0
Thr-280
O␥1
2.9
Asn-24
N␦2
2.5
O3 O31 O31 O32 O32
Lys-173 Lys-173 Arg-184 Arg-184 Lys-197
N N N⑀ N2 N
3.1 3.1 2.7 2.7 2.8
Lys-273 Lys-273 Arg-284 Arg-284 Arg-305
N N N⑀ N2 N⑀
3.4 3.5 2.9 2.9 3.4
Lys-12 Lys-12 Arg-28 Arg-28
N N N⑀ N2
3.1 2.9 3.1 3.0
O41 O41 O42 O42 O43 O43
Lys-173 Arg-235 Tyr-195 Lys-173 Arg-206 Arg-235
N N2 O N N1 N1
2.7 2.9 2.6 3.4 3.0 2.7
Lys-273 His-355 Tyr-295 Lys-273
N N⑀2 O N
2.6 2.6 2.6 3.2
Lys-12
N
2.8
Tyr-39
O
2.3
Ser-21
O␥
2.9
Asn-354 Lys-343
N␦2 N
2.8 2.6
Lys-18 Ser-21
N O␥
2.7a 2.3
O6
Ser-21
O␥
2.8
O41
Gln-15
N
2.8
Gln-16 Lys-17 Lys-17 Lys-18
N N N N
3.4 2.9 3.0 2.8
O5 O51 O52 O53
O5 O51 O51 O53 O53
Gly-176 Leu-177 Gly-176 Val-178
N N N N
3.1 2.8 3.2 3.1
Gly-276 Arg-277
N N
3.2 3.1
Arg-277 Val-278
N N
3.0 2.8
a Data for Btk-PH are based on the LIGPLOT output for Btk-PH E41K (chain A: see Figure 3 of Baraldi et al., 1999). However, measurement of the distance between the side chain of Btk-PH K18 and the Ins(1,3,4,5)P4 5-phosphate in PDB entries 1BWN and 1B55 gives 3.8 A˚–4.5 A˚ rather than 2.7 A˚ quoted here.
2, top) contributes ⵑ2.6 kcal/mol to the free energy of Ins(1,3,4,5)P4 binding to Grp1-PH, for which the Ins(1,3,4,5)P4/Ins(1,3,4)P3 binding selectivity is 85-fold (Table 1). Since the Ins(1,3,4,5)P4/Ins(1,3,4)P3 selectivity of DAPP1-PH is only 3.5-fold, it can be concluded that the 5-phosphate contributes only ⵑ0.7 kcal/mol to Ins(1,3,4,5)P4 binding to this PH domain. Comparison with the Btk PH Domain When the Ins(1,3,4,5)P4 binding site of the Btk-PH/ Ins(1,3,4,5)P4 complex structure (Baraldi et al., 1999) is compared with those of DAPP1-PH and Grp1-PH, it can be seen that they are very similar (Figure 4). All of the Ins(1,3,4,5)P4 interactions common to DAPP1-PH and Grp1-PH are also seen with Btk-PH (Figure 4C; righthand side). The Ins(1,3,4,5)P4 interactions that are unique to Btk-PH (left in Figure 4C; Table 2, top) are made exclusively with the 5-phosphate group, as expected given the Ins(1,3,4,5)P4/Ins(1,3,4)P3 selectivity for Btk-PH of 600-fold (Kojima et al., 1997). As in the Grp1PH/Ins(1,3,4,5)P4 complex, the 3-, 4-, and 5-phosphates are all 90% (or more) buried in the Btk-PH/Ins(1,3,4,5)P4 complex (Figure 5C). A clear pocket for the 5-phosphate group is seen on the Btk-PH surface. This pocket does not result from a unique insertion as seen in Grp1-PH, but from the fact that the longer 1/2 loop of Btk-PH (Figure 6A) is able to envelop the 5-phosphate and make side chain (K18, S21) as well as backbone hydrogen bonds with this group. Thus, the extended 1/2 loop
of Btk-PH plays a role in Ins(1,3,4,5)P4 binding by this PH domain that is analogous to the role of the Grp1-PH 6/7 insertion discussed above. Structural Explanation of the 3-Phosphoinositide Binding Motif Isakoff and coworkers have described a sequence motif that serves as a good predictor of high-affinity 3-phosphoinositide binding for PH domains (Isakoff et al., 1998). This motif is shown in Figure 6 and spans the 1/ 2 loop. The first lysine in the motif (half red, half green) hydrogen bonds with both the 3- and 4-phosphates of Ins(1,3,4,5)P4. Since the 1/2 loop backbone that follows this lysine comes very close to the bound Ins(1,3,4,5)P4 (Figure 3), a large side chain two residues after it (on the same side of the strand) would clash with the bound ligand. Accordingly, a small amino acid (G/A/ S/P) is present at this position in all 3-phosphoinositide binders. The length of the 1/2 loop in PH domains varies, and the motif suggests that it must contain at least one amino acid with a basic side chain. R277 of Grp1-PH and K18 of Btk-PH (Figure 6A) fulfill this requirement, and both come sufficiently close to the 5-phosphate of Ins(1,3,4,5)P4 to participate in ligand binding (Figures 4B and 4C). By contrast, the only basic residue in the DAPP1-PH 1/2 loop is K179 (Figure 6B), the side chain of which is over 10 A˚ away from Ins(1,3,4,5)P4 in the DAPP1-PH/Ins(1,3,4,5)P4 complex (Figure 4A) and is unlikely to participate in binding. The
PH Domain Recognition of 3-Phosphoinositides 379
Figure 4. Interactions that Define 3-Phosphoinositide Binding Specificities of PH Domains Ins(1,3,4,5)P4 is shown in the binding sites of (A) DAPP1-PH, (B) Grp1-PH, and (C) Btk-PH (Baraldi et al., 1999). C␣ positions of the core PH domains were overlaid as for Figure 3 and are viewed from essentially the same orientation. The 1/2 backbone is represented as a worm through the C␣ positions and is colored green. In the orientation chosen, all side chains that participate in the common set of Ins(1,3,4,5)P4 interactions (see text and Table 2, top) are on the right hand side of the Ins(1,3,4,5)P4 molecule. Unique interactions that define specificity differences (italics in Table 2, top; see text) all come from the left-hand side of the bound inositol phosphate in this orientation. Labels for residues with side chains that participate in the unique specificity-determining interactions are boxed, and color-coded according to whether they interact with the 4-phosphate (green) or the 5-phosphate (blue) of Ins(1,3,4,5)P4. The 1/2 and 3/4 loop backbones are marked, as is the region of the DAPP1-PH 6/7 loop that contributes R235, and the 6/7 insertion of Grp1-PH that includes strands 6⬘ and 6⬘⬘.
basic side chain in the 1/2 loop therefore appears to be important in PtdIns(3,4,5)P3-specific PH domains, but not necessarily in those that bind both PtdIns(3,4,5)P3 and PtdIns(3,4)P2. The next requirement of the motif is a basic residue at the beginning of 1 (K282 in Grp1PH; K182 in DAPP1-PH), which forms direct or watermediated hydrogen bonds to the 1-phosphate of bound Ins(1,3,4,5)P4 (Figure 3). Next is a conserved arginine (R284 in Grp1-PH; R184 in DAPP1-PH), which is involved in critical contacts with the 3-phosphate of the bound ligands. This is the position of the first-reported XLA-asso-
ciated mutations in Btk, which impair PtdIns(3,4,5)P3 binding by the Btk PH domain (Baraldi et al., 1999). The final position in the motif reflects a structural requirement for a hydrophobic side chain that projects into the hydrophobic core.
Predicting Binding Specificity for 3-Phosphoinositide Binding PH Domains The two key requirements for Grp1-like PtdIns(3,4,5)P3 binding specificity are: (1) that a cleft or pocket for the
Figure 5. Grp1-PH and Btk-PH Bury the 3-, 4-, and 5-Phosphates While DAPP1-PH Buries Only the 3- and 4-Phosphate GRASP surface representations (Nicholls et al., 1991) of the inositol phosphate binding sites in (A) DAPP1-PH, (B) Grp1-PH, and (C) Btk-PH (Baraldi et al., 1999), shown in the same orientations as in Figure 4. The accessible surface is colored on the basis of its electrostatic potential (⫺25kT and ⫹25kT corresponding to fully red and fully blue, respectively). The bound inositol phosphate is colored magenta, and phosphate positions are labeled P1, P3, P4, and P5. The projection of the 5-phosphate into a pocket on the surface of Grp1-PH (B) and Btk-PH (C), but not DAPP1-PH (A), is illustrated.
Molecular Cell 380
Figure 6. Prediction of PH Domain 3-Phosphoinositide Specificity PH domains shown to recognize PI 3-kinase products (Isakoff et al., 1998) are aligned with Grp1-PH and Btk-PH (A) or DAPP1-PH (B), according to whether they are predicted (see text) to make direct side chain contacts with the 5-phosphate of Ins(1,3,4,5)P4. Elements of secondary structure are delineated with gray arrows ( strands) or a black bar (the C-terminal ␣ helix). Residues are colored when their side chain is involved in interactions with Ins(1,3,4,5)P4 in the Btk-PH, Grp1-PH, or DAPP1-PH complex structures. Yellow represents interaction with the 1-phosphate; red, the 3-phosphate; green, the 4-phosphate; and blue, the 5-phosphate. Color coding is predicted for PH domains of unknown structure. The 3-phosphoinositide binding motif (Isakoff et al., 1998) in the 1/2 region is also color coded as described above. In (A), PH domains with names underlined are known to select PtdIns(3,4,5)P3 over PtdIns(3,4)P2. Others are predicted to do so. In (B), DAPP1-PH and PKB-PH are both known to bind almost equally to PtdIns(3,4,5)P3 and PtdIns(3,4)P2. Others are predicted to do so.
5-phosphate group (with hydrogen-bonding functional groups) is formed by an extended 1/2 loop, a 6/7 insertion, or some other additional sequences within the PH domain; and (2) that a positively charged side chain in the 1/2 loop (or elsewhere) imparts a basic character to this cleft or pocket. Figures 4 and 5 show how this is achieved by Grp1-PH and Btk-PH, but not DAPP1-PH. In Figure 6, the sequences of several PH domains are aligned in two categories. PH domains in category A are those known to be PtdIns(3,4,5)P3 specific (for which names are underlined) and those that we predict are PtdIns(3,4,5)P3 specific. In category B are PH domains known to bind both PtdIns(3,4,5)P3 and PtdIns(3,4)P2 (underlined), plus those that we predict bind both 3-phosphoinositides. The PH domains from DOS, Gab1, Gap1IP4BP, Gap1m, and PDK1 all maintain most of the residues that are critical for Ins(1,3,4,5)P4 binding (Figure 6A). Each of these PH domains also has a 1/2 loop similar in length to that of Btk-PH, and which contains one or more basic residues close to its beginning. We suggest that the
1/2 loops of these PH domains are likely to be capable of enveloping the 5-phosphate of Ins(1,3,4,5)P4 in a positively charged pocket, as the 1/2 loop of Btk-PH is seen to do in Figures 4C and 5C. We would predict that these PH domains are all PtdIns(3,4,5)P3 specific, as has actually been shown for the Gab1 and Gap1IP4BP PH domains (Cullen et al., 1995; Rodrigues et al., 2000). The PtdIns(3,4,5)P3 specificity of Grp1-PH depends on the 5-phosphate pocket that formed by side chains from the 6/7 insertion, and the R277 side chain in 1/2. The contribution of the 6/7 insertion in Grp1-PH substitutes for a long 1/2 loop. Based on these considerations, a PH domain that has a short 1/2 loop, no other insertion to substitute for it, and no basic residue at the beginning of the 1/2 loop should not be capable of forming the pocket seen in Figures 5B and 5C for the 5-phosphate of Ins(1,3,4,5)P4. A 3-phosphoinositide binding PH domain lacking these three characteristics is likely to bind similarly well to PtdIns(3,4,5)P3 and PtdIns(3,4)P2, as do DAPP1-PH (Table 1) and PKB-PH (Frech et al., 1997).
PH Domain Recognition of 3-Phosphoinositides 381
we predicted should resemble DAPP1-PH in its interactions, both Ins(1,3,4)P3 and Ins(1,3,4,5)P4 are seen to bind strongly (Figure 7D). Thus, at least in this case, we were able to predict correctly the inositol phosphate binding specificity of an uncharacterized PH domain.
Figure 7. Demonstration of Dual Specificity of an Uncharacterized PH Domain The PH domain encoded by accession number AA054961, predicted to bind both PtdIns(3,4,5)P3 and PtdIns(3,4)P2 in Figure 6B, binds with high affinity to the soluble head groups of both phosphoinositides (Ins(1,3,4,5)P4 and Ins(1,3,4)P3). A simple gel filtration assay is used (Lemmon et al., 1995). In (A), the elution positions of PH domain (protein) and 3H-labeled inositol phosphate, when passed through a Bio-Gel P6 column, are marked. In (B), both 3H-Ins(1,3,4,5)P4 (cyan) and 3H-Ins(1,3,4)P3 (red) coelute with DAPP1-PH, indicating highaffinity binding of both (Kd ⱕ 1M). In (C), the same experiment shows that only 3H-Ins(1,3,4,5)P4 (cyan) binds strongly to Grp1-PH. In part (D), the prediction made in Figure 6B is confirmed—the AA054961 PH domain binds strongly to both Ins(1,3,4,5)P4 (cyan) and Ins(1,3,4)P3 (red).
Conclusions To understand why Grp1-PH is highly specific for PtdIns(3,4,5)P3, while DAPP1-PH binds strongly to both PtdIns(3,4)P2 and PtdIns(3,4,5)P3, we determined the X-ray crystal structures of both PH domains in complex with Ins(1,3,4,5)P4, and compared these structures with that previously determined for the Btk-PH/Ins(1,3,4,5)P4 complex (Baraldi et al., 1999). Approximately 75% of the interactions made with Ins(1,3,4,5)P4 are common to the three different PH domains and involve only the 1-, 3-, and 4-phosphates. Side chains that participate in these common interactions overlay in the different PH domain structures with remarkable similarity. The common set of interactions does not appear to be sufficient for highaffinity binding to inositol phosphates or phosphoinositides, and the remaining 25% of the hydrogen bonding interactions required for high-affinity binding are unique to each PH domain. In DAPP1-PH, these unique interactions are all made with the 4-phosphate of Ins(1,3,4,5)P4, while in Grp1-PH and Btk-PH most are made to the 5-phosphate. Thus, the 5-phosphate group contributes very little to Ins(1,3,4,5)P4 binding by DAPP1-PH, and this PH domain can bind to both PtdIns(3,4)P2 and PtdIns(3,4,5)P3. By contrast, the 5-phosphate is critical for high-affinity Ins(1,3,4,5)P4 binding by Grp1-PH or BtkPH, which both therefore bind PtdIns(3,4,5)P3 but not PtdIns(3,4)P2. Using the sequence motif identified by Isakoff and coworkers (Isakoff et al., 1998), it is possible to predict for novel PH domains whether they bind to the lipid second messengers produced upon PI 3-kinase activation. The structural information presented here provides an explanation of this motif, and, importantly, allows us to further predict whether novel 3-phosphoinositide binding PH domains bind solely to PtdIns(3,4,5)P3, or to both PtdIns(3,4,5)P3 and PtdIns(3,4)P2. Whether agonist-dependent PH domain-mediated recruitment of a signaling protein can be driven by both of these lipid second messengers, or by only PtdIns(3,4,5)P3, is likely to have a significant impact on the nature and time course of its response. Experimental Procedures
Figure 6B aligns DAPP1-PH and PKB-PH with other PH domains expected to have similar binding specificities. To test these predictions, we generated the AA054961 PH domain as a GST fusion protein in E. coli and used a gel filtration assay (Lemmon et al., 1995) to determine whether it can bind with high affinity to Ins(1,3,4)P3 and/ or Ins(1,3,4,5)P4 (Figure 7). In this assay, tritiated inositol phosphate is mixed with the PH domain, and the mixture is run through a desalting column. If significant counts elute from the column with the protein, strong binding is indicated. By contrast, if binding is only weak, any complex formed will dissociate upon gel filtration, and all inositol phosphate counts will elute at the salt position (Figure 7A). Using this simple assay, we confirmed that both Ins(1,3,4)P3 and Ins(1,3,4,5)P4 bind strongly to DAPP1-PH (Figure 7B), while only Ins(1,3,4,5)P4 binds strongly to Grp1-PH (Figure 7C). When the same experiment is performed with the AA054961 PH domain, which
Production of Grp1-PH and DAPP1-PH DNA fragments encoding residues 264–391 of mouse Grp1 (Klarlund et al., 1997) and residues 148–273 of human DAPP1 were amplified using PCR. A unique NdeI site was included at the N-terminal methionine codon, and unique Bam HI (Grp1) or BglII (DAPP1) sites after the stop codon. Purified and digested PCR products were ligated into NdeI/BamHI-cut pET11a, and DNA sequences were confirmed using standard procedures. Grp1-PH and DAPP1-PH were expressed and purified as described for the dynamin and PLC-␦1 PH domains (Ferguson et al., 1994, 1995), with the addition of a final high-resolution cation exchange step. Selenomethionine (SeMet) protein was produced in the methionine auxotroph B834 (DE3) (Novagen), grown in MOPSbased medium (Neidhardt et al., 1974) containing 50 mg/liter D-Lselenomethionine (Sigma) as described (Ferguson et al., 1994). Induction at 20⬚C was required for soluble expression of SeMet PH domains. Purified proteins were buffer exchanged into 25 mM sodium phosphate (pH 7.0) (DAPP1-PH) or 10 mM MES (pH 6.0), 100 mM NaCl (Grp1-PH), and concentrated to 50 mg/ml. Samples for crystallization contained protein at 25 mg/ml (DAPP1-PH) or 15 mg/
0.007 A˚ 1.24⬚
aa 157–262 (879) — 2 PO42⫺ ions 32
500–2.4 A˚ 4243/740 21.9 (27.6)
0.005 A˚ 1.29⬚
aa 162–261 (838) Ins(1,3,4,5)P4 — 105
500–1.8 A˚ 10338/1083 21.9 (23.2)
DAPP1-PH, aa 162–255
X25 1.8 A˚ 50768/11260 94.8 (99.4) 0.04 (0.05) 26.6 (29.0)
P212121 a ⫽ 40.7 A˚, b ⫽ 46.3 A˚, c ⫽ 64.7 A˚
DAPP1-PH/ Ins(1,3,4,5)P4 Complex
0.006 A˚ 1.21⬚
aa A265–386, B266–391 (2012) 2 Ins(1,3,4,5,6)P5 SO42⫺ 164
500–1.9 A˚ 16563/1795 22.9 (27.0)
15–3.0 A˚ Se-Metf 0.44/0.34/0.79 Mersalyl 1.45/1.11/0.23 UO2SO4 0.48/0.39 0.49, 0.51
P21 a ⫽ 38.0 A˚, b ⫽ 72.2 A˚, c ⫽ 46.9 A˚,  ⫽ 93.2⬚ (2 Grp1-PH per a.u.i) X25 1.9 A˚ 53932/18374 95.2 (80.5) 0.05 (0.08) 20.0 (10.7)
Grp1-PH/Ins(1,3,4,5,6)P5 Complex
0.007 A˚ 1.29⬚
A264–386, B264–389 (2062) 2 Ins(1,3,4,5)P4 2 SO42⫺ ions 126
500–2.5 A˚ 13953/1458 22.8 (28.6)
Grp1-PH, aa A266–A387, B266–B387
X12B 2.5 A˚ 31494/15962 94.2 (90.2) 0.05 (0.23) 16.9 (4.9)
P212121 a ⫽ 55.9 A˚, b ⫽ 64.5 A˚, c ⫽ 69.8 A˚ (2 Grp1-PH per a.u.)
Grp1-PH/Ins(1,3,4,5)P4 Complexa
b
a
Data were collected from an Se-Met-containing crystal at ⫽ 0.98 A˚. Data collection statistics refer to a data set in which the Friedel pairs were not merged. Numbers in parentheses refer to the last resolution shell. c X25, X12B: National Synchrotron Light Source (NSLS) beamlines X25 and X12B, Brookhaven National Laboratory. d Rsym ⫽ R|Ih ⫺ ⬍Ih⬎|/RIh, where ⬍Ih⬎ is the average intensity over symmetry equivalent measurements. e Phasing power ⫽ R|FH|/R||FPHO| ⫺ |FPHC||; anomalous phasing power ⫽ R|FH″| / ||ADO| ⫺ |ADc||, where AD is the anomalous difference. f Data were collected at X-12B at ⫽ 0.98 A˚. All other derivative data were collected in-house. g Figure of merit ⫽ 兰 P (φ)exp(iφ)dφ / 兰 P(φ)dφ, where P is the probability distribution of φ, the phase angle. h R factor ⫽ R|FO ⫺ FC| / RFO, where summation is over data used in the refinement; free R factor includes only 10% of the data excluded from all refinements. i Asymmetric unit.
Bond length Bond angles
RMS deviations
Protein (number of atoms) Ligand Bound ions Bound water molecules
Model
Resolution limits No. of reflections/no. test set R factor (R free)h
Refinement Statistics
Mean FOMg acentric/centric Molecular replacement search model
Resolution limits Phasing powere acentric/centric/anom
50–2.6 A˚ HgCl2 1.91/0.66/0.49 K2PtCl4 0.98/0.91/0.22 HAuCl4 0.75/0.67/0.21 0.46/0.65
X25 2.4 A˚ 17635/4973 94.0 (81.3) 0.05 (0.14) 21.5 (5.2)
X-ray sourcec Resolution limit Observed/unique Completeness (%) Rsymd I/
Phasing Statistics
I4 a ⫽ 83.4 A˚, c ⫽ 38.4 A˚
DAPP1-PH
Space group Unique cell dimensions
Data Collection Statisticsb
Table 3. Summary of Crystallographic Data
Molecular Cell 382
PH Domain Recognition of 3-Phosphoinositides 383
ml (Grp1-PH), plus a 1.4-fold molar excess of inositol phosphate (when added). D-myo-Ins(1,3,4,5,6)P5 was from Matreya Inc., and D-myo-Ins(1,3,4,5)P4 from Sigma or Alexis Inc. Crystallization DAPP1-PH Tetragonal crystals of DAPP1-PH (I4, a ⫽ b ⫽ 83.4 A˚, c ⫽ 38.4 A˚) grew at 18⬚C from 15%–20% PEG 3450, 20%–30% ethylene glycol, 50 mM NaPO4, 200 mM (NH4)2PO4 (pH 7.0), by hanging-drop vapor diffusion. Native crystals were frozen directly from the drop. Derivatives were obtained by soaking crystals in a stabilizer of 25% PEG 3450, 20% ethylene glycol, 50 mM NaPO4, 200 mM (NH4)2PO4 plus (1) 1 mM HgCl2 (3 hr) (2) 0.01 mM HgCl2 (2 hr) (3) 1 mM K2PtCl4 (20 hr), and (4) 1 mM HAuCl4 (15 hr). DAPP1-PH/Ins(1,3,4,5)P4 Orthorhombic crystals of the DAPP1-PH/Ins(1,3,4,5)P4 complex (P212121, a ⫽ 40.7 A˚, b ⫽ 46.7 A˚, c ⫽ 64.7 A˚) were obtained at 18⬚C from 12.5% PEG 3450, 50 mM Tris (pH 7.0). 18% PEG 3450, 10% ethylene glycol, 50 mM Tris (pH 7.0) was used as cryostabilizer. Grp1-PH/Ins(1,3,4,5,6)P5 Crystals of the Grp1-PH/Ins(1,3,4,5,6)P5 complex (P21, a ⫽ 38 A˚, b ⫽ 72.2 A˚, c ⫽ 46.9 A˚,  ⫽ 93.2⬚) were obtained by microseeding at 21⬚C from 2%–10% PEG 8000, 50 mM sodium acetate, 200 mM (NH4)2SO4 (pH 5.0). Native crystals were cryostabilized in 15% PEG 8000, 22% ethylene glycol, 200 mM (NH4)2SO4 (pH 5.0). Derivatives were obtained by soaking crystals in a stabilizer of 15% PEG 8000, 200 mM (NH4)2SO4, 50 mM NaOAc (pH 5.0) plus (1) 1 mM sodium o-[(3-hydroxymercuri-2-methoxy-propyl)carbamoyl]phenoxyacetate (sodium mersalylate) (12 hr); (2) 5 mM sodium tungstate (12 hr); (3) 0.5 mM gadolinium acetate (12 hr); (4) 0.5 mM samarium acetate (12 hr) or (5) 2 mM uranyl sulfate (12 hr), and were cryostabilized as for native crystals. Selenomethionyl Grp1-PH/Ins(1,3,4,5,6)P5 crystals were briefly exposed to a cryostabilizer of 30% PEG 8000, 50 mM sodium acetate, 200 mM (NH4)2SO4 (pH 5.0) before freezing. Grp1-PH/Ins(1,3,4,5)P4 Crystals of selenomethionyl Grp1-PH/Ins(1,3,4,5)P4 complex (P212121, a ⫽ 55.9 A˚, b ⫽ 64.5 A˚, c ⫽ 69.8 A˚) were obtained from 5% PEG 8000, 50 mM sodium acetate, 200 mM (NH4)2SO4 (pH 5.0). Crystals were briefly exposed to a cryostabilizer of 15% PEG 8000, 22% ethylene glycol, 200 mM (NH4)2SO4, 50 mM sodium acetate (pH 5.0), before freezing. Data Collection All cryostabilized crystals were flash frozen in liquid nitrogen. Data were collected at 100K at beamline X25 or beamline X12B of the National Synchrotron Light Source (NSLS), using Brandeis B4 CCD detectors, or in house using a MAR image plate detector and CuK␣ X rays. Data were processed using DENZO and SCALEPACK (Otwinowski and Minor, 1994). A summary of all statistics for data collection, phasing, and crystallographic refinement is given in Table 3. Structure Determination and Refinement DAPP1-PH The three heavy-atom derivatives of DAPP1-PH share a common heavy-atom site, identified in each case in the isomorphous and anomalous difference Patterson maps. The HgCl2 derivative has a second unique site identified in Patterson and difference Fourier maps. Heavy atom sites were refined, and experimental phases determined, using a locally modified version of MLPHARE (Otwinowski, 1991). Phases were improved by solvent flattening (Abrahams and Leslie, 1996) and partial model combination using a local density modification program DPHASE (Greg Van Duyne). The program O (Jones et al., 1991) was used for model building. The structure was refined against all data to 2.4 A˚ using CNS (Bru¨nger et al., 1998) implementing bulk solvent correction. DAPP1-PH/Ins(1,3,4,5)P4 The coordinates for amino acids 162–255 of the refined unliganded DAPP1-PH structure were used as a search model to solve the DAPP1-PH/Ins(1,3,4,5)P4 structure by molecular replacement (MR) with AmoRe (Navaza and Saludjian, 1997). Clear density for the Ins(1,3,4,5)P4 could be seen in the initial MR-phased electron density map. After an initial round of manual rebuilding and refinement, the ligand was included in the model. The structure was refined against
all data to 1.8 A˚ as described above. Composite simulated-annealing omit maps were used in rebuilding of the Ins(1,3,4,5)P4 binding site and portions of the PH domain involved in crystal packing interactions. Grp1-PH/Ins(1,3,4,5,6)P5 Three distinct heavy-atom positions for the mercury (mersalyl) and selenomethionine-containing derivatives were identified in isomorphous and anomalous difference Patterson maps, respectively. An additional minor heavy-atom site was identified for each of these derivatives in difference Fourier maps. Additional phase information came from weakly occupied heavy-atom sites (identified in difference Fourier maps) in a uranyl, a gadolinium, a samarium, and a tungstate derivative. Map calculations, model building, and refinement (to 1.9 A˚) were performed as for DAPP1-PH. The two molecules in the asymmetric unit were built and refined independently. Grp1-PH/Ins(1,3,4,5)P4 The coordinates (protein only) for amino acids 266–387 of molecule A from the refined Grp1-PH/Ins(1,3,4,5,6)P5 structure were used as a search model to solve the Grp1-PH/Ins(1,3,4,5)P4 complex structure by MR using AmoRe (Navaza and Saludjian, 1997). Two solutions were obtained, as expected for the two molecules in the asymmetric unit. Examination of the packing of these solutions revealed that the arrangement of the two molecules in the asymmetric unit was identical to that in the Grp1-PH/Ins(1,3,4,5,6)P5 complex. Clear density for the two Ins(1,3,4,5)P4 molecules was seen in the initial MR-phased electron density map. MR phases were used to locate the positions of the selenium atoms (selenomethionine protein was used to generate these crystals). AS phase information was combined with the MR phases to reduce model bias. Model building and refinement were performed as for the DAPP1-PH/Ins(1,3,4,5)P4 complex. Unless otherwise noted, structure figures were generated with MOLSCRIPT (Kraulis, 1991) and Raster3D (Merritt and Bacon, 1997). Acknowledgments We are sincerely grateful to Professor Greg Van Duyne for his generosity with crystallographic advice and equipment, and many valuable discussions and comments on the manuscript. We also thank Malcolm Capel and Hal Lewis of the National Synchrotron Light Source (NSLS) at Brookhaven National Laboratory for help in data collection. The NSLS is supported by the US Department of Energy, Division of Materials Sciences and Division of Chemical Sciences, under contract number DE-AC02-98CH10886. X25 is supported in part by the National Institutes of Health (NIH), the National Center for Research Resources, and the US Department of Energy, Office of Biological and Environmental Research. These projects were supported by the NIH (NIGMS grant GM56846 to M. A. L.; NIDDK grant DK49207 to E. Y. S.), the US Army Breast Cancer Research Program (DAMD17-98-1-8228 to K. M. F.), and a Damon Runyon Scholar Award from the Cancer Research Fund of the Damon Runyon-Walter Winchell Foundation (to M. A. L.). Received April 4, 2000; revised May 30, 2000. References Abrahams, J.P., and Leslie, A.G.W. (1996). Methods used in the structure determination of bovine mitochondrial F1 ATPase. Acta Crystallogr. D. 52, 30–42. Banfic, H., Tang, X.-W., Batty, I.H., Downes, C.P., Chen, C.-S., and Rittenhouse, S.E. (1998). A novel integrin-activated pathway forms PKB/Akt-stimulatory phosphatidylinositol 3,4-bisphosphate via phosphatidylinositol 3-phosphate in platelets. J. Biol. Chem. 273, 13–16. Baraldi, E., Djinovic Carugo, K., Hyvo¨nen, M., Lo Surdo, P., Riley, A.M., Potter, B.V.L., O’Brien, R., Ladbury, J.E., and Saraste, M. (1999). Structure of the PH domain from Bruton’s tyrosine kinase in complex with inositol 1,3,4,5-tetrakisphosphate. Structure 7, 449–460. Bolland, S., Pearse, R.N., Kurosaki, T., and Ravetch, J.V. (1998). SHIP modulates immune receptor responses by regulating membrane association of Btk. Immunity 8, 509–516.
Molecular Cell 384
Bru¨nger, A.T., Adams, P.D., Clore, G.M., Gros, P., Grosse-Kuntsleve, R.W., Jiang, J.-S., Kuszewski, J., Nilges, M., Pannu, N.S., and Read, R.J. (1998). Crystallography and NMR system: a new software system for macromolecular structure determination. Acta Crystallogr. D 54, 905–921.
Marshall, C.J. (1995). Specificity of receptor tyrosine kinase signaling: transient versus sustained extracellular signal-regulated kinase activation. Cell 80, 179–185.
Cullen, P.J., Chung, S.K., Chang, Y.T., Dawson, A.P., and Irvine, R.F. (1995). Specificity of the purified inositol (1,3,4,5) tetrakisphosphatebinding protein from porcine platelets. FEBS Lett. 358, 240–242.
Navaza, J., and Saludjian, P. (1997). AMoRe: an automated molecular replacement program package. Methods Enzymol. 276, 581–594.
Damen, J.E., Liu, L., Rosten, P., Humphries, R.K., Jefferson, A.B., Majerus, P.W., and Krystal, G. (1996). The 145-kDa protein induced to associate with Shc by multiple cytokines is an inositol tetraphosphate and phosphatidylinositol 3,4,5-triphosphate 5-phosphatase. Proc. Natl. Acad. Sci. USA 93, 1689–1693. Dowler, S., and Currie, R.A. Downes, C.P., and Alessi, D.R. (1999). DAPP1: a dual adaptor for phosphotyrosine and 3-phosphoinositides. Biochem. J. 342, 7–12. Falasca, M., Logan, S., Lehto, V.P., Baccante, G., Lemmon, M.A., and Schlessinger, J. (1998). Activation of phospholipase C␥ by PI 3-kinase induced PH domain mediated membrane targeting. EMBO J. 17, 414–422. Ferguson, K.M., Lemmon, M.A., Schlessinger, J., and Sigler, P.B. (1994). Crystal structure at 2.2A˚ of the pleckstrin homology domain from human dynamin. Cell 79, 199–209. Ferguson, K.M., Lemmon, M.A., Schlessinger, J., and Sigler, P.B. (1995). Structure of a high affinity complex between inositol-1,4,5trisphosphate and a phospholipase C pleckstrin homology domain. Cell 83, 1037–1046. Franke, T.F., Kaplan, D.R., Cantley, L.C., and Toker, A. (1997). Direct regulation of the akt proto-oncogene product by phosphatidylinositol-3,4-bisphosphate. Science 275, 665–668. Frech, M., Andjelkovic, M., Ingley, E., Reddy, K.K., Falck, J.R., and Hemmings, B.A. (1997). High affinity binding of inositol phosphates and phosphoinositides to the pleckstrin homology domain of RAC/ protein kinase B and their influence on kinase activity. J. Biol. Chem. 272, 8474–8481. Gray, A., Van der Kaay, J., and Downes, C.P. (1999). The pleckstrin homology domains of protein kinase B and GRP1 (general receptor for phosphoinositides-1) are sensitive and selective probes for the cellular detection of phosphatidylinositol 3,4-bisphosphate and/or phosphatidylinositol 3,4,5-trisphosphate in vivo. Biochem. J. 344, 929–936. Isakoff, S.J., Cardozo, T., Andreev, J., Li, Z., Ferguson, K.M., Abagyan, R., Lemmon, M.A., Aronheim, A., and Skolnik, E.Y. (1998). Identification and analysis of PH domain-containing targets of phosphatidylinositol 3-kinase using a novel in vivo assay in yeast. EMBO J. 17, 5374–5387. Jones, T.A., Zou, J.Y., Cowan, S.W., and Kjeldgaard, M. (1991). Improved methods for binding protein models in electron density maps and the location of errors in these models. Acta. Crystallogr. A. 47, 110–119. Kavran, J.M., Klein, D.E., Lee, A., Falasca, M., Isakoff, S.J., Skolnik, E.Y., and Lemmon, M.A. (1998). Specificity and promiscuity in phosphoinositide binding by pleckstrin homology domains. J. Biol. Chem. 273, 30497–30508. Klarlund, J.K., Guilherme, A., Holik, J.J., Virbasius, A., and Czech, M.P. (1997). Signaling by 3,4,5-phosphoinositide through proteins containing pleckstrin and Sec7 homology domains. Science 275, 1927–1930. Kojima, T., Fukuda, M., Watanabe, Y., Hamazato, F., and Mikoshiba, K. (1997). Characterization of the pleckstrin homology domain of Btk as an inositol polyphosphate and phosphoinositide binding domain. Biochem. Biophys. Res. Comm. 236, 333–339. Kraulis, P.J. (1991). MOLSCRIPT: a program to produce both detailed and schematic plots of protein structures. J. Appl. Crystallogr. 24, 946–950. Lemmon, M.A., and Ferguson, K.M. (1998). Pleckstrin homology domains. Curr. Top. Microbiol. Immunol. 228, 39–74. Lemmon, M.A., Ferguson, K.M., O’Brien, R., Sigler, P.B., and Schlessinger, J. (1995). Specific and high-affinity binding of inositol phosphates to an isolated pleckstrin homology domain. Proc. Natl. Acad. Sci. USA 92, 10472–10476.
Merritt, E.A., and Bacon, D.J. (1997). Raster3D: photorealistic molecular graphics. Methods Enzymol. 277, 505–524.
Neidhardt, F.C., Bloch, P.L., and Smith, D.F. (1974). Culture medium for enterobacteria. J. Bacteriol. 119, 736–747. Nicholls, A., Sharp, K.A., and Honig, B. (1991). Protein folding and association: insights from the interfacial and thermodynamic properties of hydrocarbons. Proteins 11, 281–296. Otwinowski, Z. (1991). MLPHARE. In CCP4 Proceedings (Warrington, UK: Daresbury Laboratory), 80–88. Otwinowski, Z., and Minor, W. (1994). Processing of X-ray diffraction data in oscillation mode. Methods Enzymol. 276, 307–326. Rao, V.R., Corradetti, M.N., Chen, J., Peng, J., Yuan, J., Prestwich, G.D., and Brugge, J.S. (1999). Expression cloning of protein targets for 3-phosphorylated phosphoinositides. J. Biol. Chem. 274, 37893– 37900. Rebecchi, M.J., and Scarlata, S. (1998). Pleckstrin homology domains: a common fold with diverse functions. Annu. Rev. Biophys. Biomol. Struct. 27, 503–528. Rodrigues, G.A., Falasca, M., Zhang, Z., Ong, S.H., and Schlessinger, J. (2000). A novel positive feedback loop mediated by the docking protein Gab1 and phosphatidylinositol 3-kinase in epidermal growth factor receptor signaling. Mol. Cell. Biol. 20, 1448–1459. Scharenberg, A.M., El-Hillal, O., Fruman, D.A., Beitz, L.O., Li, Z., Lin, S., Gout, I., Cantley, L.C., Rawlings, D.J., and Kinet, J.P. (1998). Phosphatidylinositol-3,4,5-trisphosphate (PtdIns-3,4,5-P3)/Tec kinase-dependent calcium signaling pathway: a target for SHIP-mediated inhibitory signals. EMBO J. 17, 1961–1972. Stephens, L.R., Jackson, T.R., and Hawkins, P.T. (1993). Agoniststimulated synthesis of phosphatidylinositol (3,4,5)-trisphosphate: a new intracellular signalling system? Biochim. Biophys. Acta. 1179, 27–75. Van der Kaay, J., Beck, M., Gray, A., and Downes, C.P. (1999). Distinct phosphatidylinositol 3-kinase lipid products accumulate upon oxidative and osmotic stress and lead to different cellular responses. J. Biol. Chem. 274, 35963–35968. Vanhaesebroeck, B., and Waterfield, M.D. (1999). Signaling by distinct classes of phosphoinositide 3-kinases. Exp. Cell Res. 253, 239–254. Varnai, P., Rother, K.I., and Balla, T. (1999). Phosphatidylinositol 3-kinase-dependent membrane association of the Bruton’s tyrosine kinase pleckstrin homology domain visualized in single living cells. J. Biol. Chem. 274, 10983–10989. Venkateswarlu, K., Gunn-Moore, F., Oatey, P.B., Tavare, J.M., and Cullen, P.J. (1998). Nerve growth factor- and epidermal growth factor-stimulated translocation of the ADP-ribosylation factor-exchange factor GRP1 to the plasma membrane of PC12 cells requires activation of phosphatidylinositol 3-kinase and the GRP1 pleckstrin homology domain. Biochem. J. 335, 139–146. Watton, S.J., and Downward, J. (1999). Akt/PKB localisation and 3⬘ phosphoinositide generation at sites of epithelial cell-matrix and cell-cell interaction. Curr. Biol. 9, 433–436. Yoon, H.S., Hajduk, P.J., Petros, A.M., Olejniczak, E.T., Meadows, R.P., and Fesik, S.W. (1994). Solution structure of a pleckstrinhomology domain. Nature 369, 672–675. Protein Data Bank Accession Codes Accession codes for the coordinates of the structures reported in this article are 1FB8 (DAPP1-PH), 1FAO (DAPP1-PH/Ins(1,3,4,5)P4 complex), 1FHW (Grp1-PH/Ins(1,3,4,5,6)P5 complex), and 1FHX (Grp1-PH/Ins(1,3,4,5)P4 complex).
